https://github.com/pathak-ashutosh/plant-seeds-classification
I took the Plant Seedlings dataset from Kaggle and made a classifier using a pretrained resnet-50 model and mxnet framework
https://github.com/pathak-ashutosh/plant-seeds-classification
classification cnn deep-learning fine-tuning kaggle machine-learning mxnet neural-networks python resnet-50
Last synced: about 2 months ago
JSON representation
I took the Plant Seedlings dataset from Kaggle and made a classifier using a pretrained resnet-50 model and mxnet framework
- Host: GitHub
- URL: https://github.com/pathak-ashutosh/plant-seeds-classification
- Owner: pathak-ashutosh
- License: mit
- Created: 2018-06-28T18:11:29.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2018-07-23T13:19:01.000Z (about 7 years ago)
- Last Synced: 2025-02-21T18:44:43.070Z (8 months ago)
- Topics: classification, cnn, deep-learning, fine-tuning, kaggle, machine-learning, mxnet, neural-networks, python, resnet-50
- Language: Jupyter Notebook
- Homepage:
- Size: 146 KB
- Stars: 3
- Watchers: 0
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# plant-seeds-classification
Can you differentiate a weed from a crop seedling?
The ability to do so effectively can mean better crop yields and better stewardship of the environment.
The Aarhus University Signal Processing group, in collaboration with University of Southern Denmark, has recently released a [dataset](https://vision.eng.au.dk/plant-seedlings-dataset/) containing images of approximately 960 unique plants belonging to 12 species at several growth stages.
_This project is one of my very first projects on Machine Learning. What I've done here is, I took Kaggle's ["Plant seedlings classification" dataset](https://www.kaggle.com/c/plant-seedlings-classification/data) and used [mxnet](https://mxnet.apache.org/) framework on a [pre-trained resnet-50 model](http://data.mxnet.io/mxnet/models/imagenet-11k-place365-ch/) to get highest possible performance in least possible (dev) time._
## About the data
You are provided with a training set and a test set of images of plant seedlings at various stages of growth. Each image has a filename that is its unique id. The dataset comprises 12 plant species. The goal is to create a classifier capable of determining a plant's species from a photo. The list of species is as follows:
```
Black-grass
Charlock
Cleavers
Common Chickweed
Common wheat
Fat Hen
Loose Silky-bent
Maize
Scentless Mayweed
Shepherds Purse
Small-flowered Cranesbill
Sugar beet
```
## Setting-up
### Setting directories
* Clone/download this repository.
* Make a 'data' folder in the 'plant-seeds-classification' folder.
* Open the data folder and make two more folders - 'train' and 'test'.
* Download the dataset from [here](https://www.kaggle.com/c/plant-seedlings-classification/data).
* Extract both train.zip and test.zip into their respective folders which we made above.
* Download the resnet-50 model along with it's symbols file from [here](http://data.mxnet.io/mxnet/models/imagenet-11k-place365-ch/) (this can work with any model, not just resnet, but you'll have to change some things in the jupyter notebook accordingly).
* Save the model in 'plant-seeds-classification' folder.
### Packages required:
* OpenCV
* TensorFlow
* mxnet
### Installation:
* Open the terminal/command prompt and cd to the 'plant-seeds-classification' directory.
* From your terminal, run the following
```
python im2rec.py --list --recursive train-lst data/train/
python im2rec.py --list --recursive test-lst data/test/
python im2rec.py --resize 256 --quality 100 --num-thread 16 test-lst data/test/
python im2rec.py --resize 256 --quality 100 --num-thread 16 train-lst data/train/
```
(This will create two list files and two rec files, one for both train data and test data)
* After running these commands successfully, open the .ipynb file using jupyter notebook and run the cells.
##### This project is licenced under the MIT Licence - see the [LICENCE](https://github.com/m3rc3n4ry/plant-seeds-classification/blob/master/LICENSE) file for more details.
If you find some problem running this model, feel free to raise an issue.